| The Web has become an excellent source for gathering consumer opinions. There are now numerous Web sites containing such opinions, e.g., customer reviews of products, Internet forums, discussion groups, and web blogs. However, the sheer size of opinion sources on the Web and the ambiguous nature of natural language can make it difficult for the task of analyzing such information. There has been a swell of interest to provide tools to automatically identify and extract attitudes, opinions, and sentiment in text.; This dissertation proposes a framework to analyzing and summarizing customer reviews of products. By analyzing reviews, we mean to extract features of products that have been commented by reviewers and to determine whether the opinions are positive or negative. The framework, which performs a semi-structured feature-based opinion summarization, is different task from traditional text summarization as we do not summarize the reviews by selecting a subset or rewrite some of the original sentences from the reviews to capture the main points as in the classic text summarization. Our task is performed in three steps: (1) extracting product features that have been commented on by customers; (2) identifying opinion sentences in each review and perform sentiment classification on opinion sentences; (3) aggregating and summarizing the opinions for each product feature. We also describe a system called Opinion Observer that makes use of the review analysis results to visually summarize and compare consumer opinions on different products. Experimental result shows the proposed feature based summarization framework is a very promising system. |
